IEEE ELECTRON DEVICE LETTERS, VOL. 34, NO. 2, FEBRUARY 2013
181
Enhanced Inversion Mobility on 4H-SiC (1120) Using Phosphorus and Nitrogen Interface Passivation Gang Liu, Student Member, IEEE, Ayayi C. Ahyi, Yi Xu, Tamara Isaacs-Smith, Yogesh K. Sharma, John R. Williams, Leonard C. Feldman, Senior Member, IEEE, and Sarit Dhar, Member, IEEE
Abstract—Low interface trap density and high channel mobility on nonpolar faces of 4H-SiC, such as the (1120) a-face, are of fundamental importance in the understanding of SiC MOS devices. It is also critical for high-voltage trench power MOSFET development. We report new results on the passivation of the SiO2 /a-face 4H-SiC interface using phosphorus, yielding field effect mobility of ∼125 cm2 /V · s. We also revisit the conventional NO passivation, for which a mobility of ∼85 cm2 /V · s was achieved on the a-face. These results not only establish new levels of mobility in SiC MOSFETS but also lead to further insights into factors currently limiting SiC inversion layer mobility.
observed. We also revisit the effect of NO on the a-face where an impressive μFE ∼ 85 cm2 /V · s is achieved. We note that the a-face is of significant technical importance. Development of high-performance trench power MOSFETs is extremely desirable for next-generation SiC power MOSFETs [9]. In these devices, the inversion channel is located on the sidewall of trenches, i.e., nonpolar faces such as the a-face. Thus, effective interface passivation for high mobility on the a-face is critical for the development of trench MOSFETs.
Index Terms—4H-SiC MOSFET, mobility, counter-doping.
II. FABRICATION
I. I NTRODUCTION
I
MPROVEMENT in the quality of the SiO2 / 4H-SiC interface is vital for the development of efficient SiC metal– oxide–semiconductor (MOS) technologies. Currently, the 4HSi-face (0001) is the conventional crystal face for MOSFET fabrication. For the Si-face, nitric oxide (NO) postoxidation annealing is the most established interface passivation process, which results in a field effect mobility μFE of ∼40 cm2 /V · s. Most recently, it has been shown that a phosphorus passivation process can result in a higher inversion layer mobility of ∼75−100 cm2 /V · s [1], [2]. In general, the Si-face mobility increases monotonically with a decreasing band-edge interface trap density (Dit ) [2], [3], which seems to result from a modification or passivation of the surface, although the precise mechanisms are not yet satisfactorily understood [4]. In general, inversion channel mobility on the a-face is higher than the Si-face for the same processing conditions [5]–[8]. We report results for a-face 4H-SiC MOSFETs using these new passivation schemes. A significant phosphorus-induced channel mobility enhancement with μFE ∼ 125 cm2 /V · s is
Manuscript received October 29, 2012; revised December 5, 2012; accepted December 7, 2012. Date of publication January 9, 2013; date of current version January 23, 2013. This work was supported in part by the U.S. Army Research Laboratory (W911NF-07-2-0046), by the U.S. National Science Foundation (MR-0907385), and by the II-VI Foundation Block-Gift Program. The review of this letter was arranged by Editor S.-H. Ryu. G. Liu is with the Department of Electrical and Computer Engineering, Rutgers University, Piscataway, NJ 08854 USA (e-mail:
[email protected]). A. C. Ahyi, T. Isaacs-Smith, Y. K. Sharma, J. R. Williams, and S. Dhar are with the Department of Physics, Auburn University, Auburn, AL 36849 USA. Y. Xu is with the Department of Chemistry and Chemical Biology, Rutgers University, Piscataway, NJ 08854 USA. L. C. Feldman is with the Institute for Advanced Materials, Devices and Nanotechnology, Rutgers University, Piscataway, NJ 08854 USA. Color versions of one or more of the figures in this letter are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/LED.2012.2233458
Si-face (4◦ off-axis) and a-face (on-axis) n-type 4H-SiC samples with ∼10-μm epitaxial layers doped with nitrogen at ∼1 × 1016 cm−3 were used to fabricate MOS capacitors for Dit measurements. Following a standard RCA cleaning process, samples underwent dry oxidation at 1150 ◦ C for different times depending on crystal orientation. For phosphorus passivation, a planar diffusion source (PDS) anneal was used for 4 h at 1000 ◦ C, converting the oxide layer to a phosphosilicate glass (PSG). Such a process applied to the Si-face results in high mobility, but also severe device instability under high temperature bias stress [2]. Focusing on this instability issue, we have recently presented a new thin PSG (∼10 nm) interfacial layer process that can significantly improve the stability [10] and retain the beneficial P effect. The focus of this letter is to investigate the mechanisms by which N and P result in mobility enhancement on different crystal faces of 4H-SiC, independent of the dopant introduction process. For nitridation, samples were subjected to NO annealing at 1175 ◦ C for 2 h. The PSG and NO processes both increase oxide thickness by different amounts, resulting in a PSG gate oxide of 84 nm and an NO oxide of 56 nm. Gate metallization was performed by sputtering or evaporating molybdenum. Trap density measurements were performed using the simultaneous high-frequency (100 kHz)low-frequency CV technique at room temperature. Long channel (150 μm) lateral test MOSFETs with a width of 290 μm were fabricated on a-face p-type epitaxial layers (Al doping ∼1 × 1016 cm−3 ), using the same gate oxidation and passivation procedures described above. The source and drain regions were formed using nitrogen implantation at 700 ◦ C and activated at 1550 ◦ C for 30 min in Ar, with the surface protected by a graphite cap. Source/drain ohmic contacts were formed by evaporation of Al. Room temperature μFE was extracted from measurements of drain current as a function of gate voltage for a fixed drain voltage (0.1 V).
0741-3106/$31.00 © 2013 IEEE
182
IEEE ELECTRON DEVICE LETTERS, VOL. 34, NO. 2, FEBRUARY 2013
TABLE I S URFACE C OVERAGE OF N ITROGEN OR P HOSPHORUS
etching (see Table I), indicating a different chemical bond structure than N or P simply included within the oxide [12], [13]. These species are possibly associated with the surface passivation and may even slightly penetrate into the bulk. Moreover, other methods without oxide etching (MEIS and SIMS) show that almost all N and P are retained after etching. Fig. 1. Dit versus energy 0.2–0.6 eV below the conduction band (EC ), extracted from Hi–Low C−V measurements. Circles are a-face samples, whereas triangles are on Si-face. Unfilled points are unpassivated, filled red are NO-annealed, and filled blue are PSG-annealed devices.
Fig. 2. Field effect mobility, μFE , of n-channel MOSFETs made on (circles) a-face and (triangles) Si-face, with (blue filled) PSG and (red filled) NO anneal, respectively. Si-face results are from [2].
III. R ESULTS Converting the thermally grown oxide to PSG reduces Dit on the a-face from 1.3 × 1013 cm−2 · eV−1 to ∼6 × 1011 cm−2 · eV−1 at 0.2 eV below the conduction band of 4H-SiC (see Fig. 1). (Although the extraction of absolute Dit values via C−V have been questioned [11], we only report comparisons of different processes.) The a-face PSG Dit value is very similar to that of the NO-annealed Si-face and a-face, whereas the Siface PSG has the lowest Dit . The results are consistent with previously published reports [2]. The a-face inversion channel mobility, μFE , is ∼125 cm2 / V · s with PSG and ∼85 cm2 /V · s with NO anneal, both significantly higher than the recently reported mobility on Si-face with the same corresponding passivation methods (see Fig. 2). For comparison, μFE without passivation on the Si-face is < 10 cm2 /V · s, and on the a-face ∼28 cm2 /V · s [5]. Threshold voltages Vth for a-face PSG, NO and Si-face PSG, NO are 3.6 V, 4.1 V and 1.2 V, 3.2 V, respectively. Interfacial chemical bonding on both faces with the two different passivations was studied by X-ray photoelectron spectroscopy (XPS) on the n-type capacitor samples with the gate metal removed and the oxides completely etched off by buffered oxide etch (BOE). Both in the case of P and N, a measurable amount of the passivating agent is retained after
IV. D ISCUSSION For the Si-face, Dit appears to be the limiting factor for the channel mobility, at least at low gate voltages. For NO, the mobility enhancement scales with the interfacial N content until the reduction of Dit saturates, corresponding to the saturation of the interfacial N coverage at ∼1/2 monolayer areal density [3]. This μFE −Dit correlation continues for the Si-face-PSG, where the lower Dit produced by P passivation leads to still higher mobility compared with NO annealing. On the a-face, there is no longer a strong correlation between Dit and μFE , for Dit < 5 × 1011 cm−2 , as demonstrated by the following comparisons: 1) “a-face-PSG relative to Si-facePSG”: greater Dit but higher μFE ; 2) “a-face-NO relative to a-face-PSG”: similar Dit but a large μFE difference. A signature of Dit limiting mobility is a positive mobilitytemperature coefficient, i.e., mobility increases with temperature, as seen in Si-face NO devices [14], contrary to the conventional negative coefficient in Si MOSFETs. A temperature dependence test on the a-face NO device showed a negative coefficient, indicating that Dit may no longer be the dominant limiting factor at this stage, consistent with the systematics described above. We also note that at medium doping levels, the bulk mobility applicable to the a-face is only about 10% higher than that for the Si-face [15], roughly independent of orientation within the a-face. Thus, the a-face/Si-face difference is not due to the mobility anisotropy in 4H-SiC. These results raise the following questions: I) What is the mechanism by which N and P increase the mobility? II) Why does the a-face yield consistently higher mobility than the Si-face? III) Why does P (at apparently lower coverage) result in higher mobility than N? I) Interfacial counter-doping into n-type: could answer. The reduction of Dit suggests that both N and P play a defect passivation role. In addition, they are also n-type donors in SiC, and it is possible that a small amount of each species become incorporated in a very thin layer of SiC [16] (< 2 nm for NO and PSG anneals, with the depth limit set by the XPS electron escape length and secondary ion mass spectrometry) at the interface, converting the doping type from p to n. Such process may be thought of as a self-limited doping mechanism. In addition, this thin layer can be easily depleted by the adjacent p-well without gate bias. In inversion mode, the positive charges in the n-type depletion layer will cancel part of the negative electrical field built up by the negative charges
LIU et al.: INVERSION MOBILITY ON 4H-SiC (1120) USING INTERFACE PASSIVATION
183
both by passivation (reducing Dit ) and by interfacial counter doping (reducing vertical field), with the latter mechanism more effective on the a-face than the Si-face. R EFERENCES
Fig. 3. Energy-band diagrams of an n-channel MOSFET, in (a) depletion and (b) strong inversion, where the standard enhancement mode structure is illustrated in black and the n-type counter-doping effect is highlighted in red.
in the p-well depletion region, reducing the slope of the potential drop toward the interface, effectively raising the surface potential [see Fig. 3(a)]. As VG increases to produce strong inversion, the thin n-type layer is also filled by electrons and becomes neutrally charged, which, in turn, widens the bottom of the conducting channel [see Fig. 3(b)]. As a result, for the same inversion electron density, the electric field is lower, relieving surface roughness scattering, and resulting in better mobility. Moreover, at any VG , particularly in the low field range, there are more electrons, leading to higher screening, which also improves mobility. The reduction of Vth with higher peak mobility on the same face also implies an interfacial counter-doping effect. Bulk counter-doping by N implantation has been reported [17]–[19], where normally off devices are difficult to achieve, and the channel mobility takes a high peak only at low gate voltage and sharply drops as the channel vertical electric field approaches that of normal MOSFETs. The interfacial counter-doping described above gives distinctly different results. II) Why is the a-face mobility higher than Si-face? In the higher field region, this may be related to miscut-induced roughness-related scattering, given that a-face samples are on-axis without miscut, whereas Si-face samples are miscut 4◦ off-axis toward (1120). In addition, the a-face has better lattice recovery and lower sheet resistance after implantation and activation than the Si-face [20]. The greater reactivity of this surface suggests that interfacial N and P introduced by NO or PSG anneals may become electrically active more readily on the a-face than on the Si-face, leading to a higher n-type doping level in the channel region and better mobility. III) Why on both faces does P passivation result in higher mobility than N passivation? Relative to N, P can be activated by lower anneal temperatures, even by oxidation processes [21], leading to higher n-type doping for PSG than for NO, despite a lower amount of P chemically bonded to the interface than N. V. C ONCLUSION We report new results for MOSFETs fabricated on the a-face of 4H-SiC. A significant phosphorus-induced channel mobility enhancement of ∼125 cm2 /V · s is achieved. We also revisit the effect of NO on the a-face of 4H-SiC, with an impressive μFE of ∼85 cm2 /V · s. Interface passivation effects are studied by electrical measurements in conjunction with XPS. Our results may indicate that N and P improve the interface
[1] D. Okamoto, H. Yano, K. Hirata, T. Hatayama, and T. Fuyuki, “Improved inversion channel mobility in 4H-SiC MOSFETs on Si face utilizing phosphorus-doped gate oxide,” IEEE Electron Device Lett., vol. 31, no. 7, pp. 710–712, Jul. 2010. [2] Y. K. Sharma, A. C. Ahyi, T. Issacs-Smith, X. Shen, S. T. Pantelides, X. Zhu, J. Rozen, J. R. Williams, and L. C. Feldman, “Phosphorous passivation of the SiO2 /4HSiC interface,” Solid State Electron., vol. 68, pp. 103–107, 2012. [3] J. Rozen, A. C. Ahyi, X. Zhu, J. R. Williams, and L. C. Feldman, “Scaling between channel mobility and interface state density in SiC MOSFETs,” IEEE Trans. Electron Devices, vol. 58, no. 11, pp. 3808–3811, Nov. 2011. [4] S. Wang, S. Dhar, S. R. Wang, A. C. Ahyi, A. Franceschetti, J. R. Williams, L. C. Feldman, and S. T. Pantelides, “Bonding at the SiC-SiO2 interface and the effects of nitrogen and hydrogen,” Phys. Rev. Lett., vol. 98, no. 2, p. 026101, Jan. 2007. [5] H. Yano, T. Hirao, T. Kimoto, H. Matsunami, K. Asano, and Y. Sugawara, “High channel mobility in inversion layers of 4H-SiC MOSFETs by utilizing (112~0) face,” IEEE Electron Device Lett., vol. 20, no. 12, pp. 611– 613, Dec. 1999. [6] J. Senzaki, K. Kojima, S. Harada, R. Kosugi, S. Suzuki, T. Suzuki, and K. Fukuda, “Excellent effects of hydrogen postoxidation annealing on inversion channel mobility of 4H-SiC MOSFET fabricated on (11-2 0) face,” IEEE Electron Device Lett., vol. 23, no. 1, pp. 13–15, Jan. 2002. [7] T. Kimoto, Y. Kanzaki, M. Noborio, H. Kawano, and H. Matsunami, “Interface properties of metal–oxide–semiconductor structures on 4HSiC {0001} and (11-20) formed by N2 O oxidation,” Jpn. J. Appl. Phys., vol. 44, no. 3, pp. 1213–1218, 2005. [8] S. Dhar, S. Wang, A. C. Ahyi, T. Isaacs-Smith, S. T. Pantelides, J. R. Williams, and L. C. Feldman, “Nitrogen and hydrogen induced trap passivation at the SiO2 /4H-SiC interface,” Mater. Sci. Forum, vol. 527–529, pp. 949–954, 2006. [9] B. J. Baliga, Advanced Power MOSFET Concepts. New York: SpringerVerlag, 2010, ch. 3. [10] Y. K. Sharma, A. C. Ahyi, T. Issacs-Smith, A. Modic, M. Park, Y. Xu, E. Garfunkel, L. C. Feldman, S. Dhar, and J. R. Williams, “High-mobility stable 4H-SiC MOSFETs using a thin PSG interfacial passivation layer,” IEEE Electron Device Lett., vol. 34, no. 2, pp. 175–177, Feb. 2013. [11] H. Yoshioka, T. Nakamura, and T. Kimoto, “Generation of very fast states by nitridation of the SiO2 /SiC interface,” J. Appl. Phys., vol. 112, no. 2, p. 024520, Jul. 2012. [12] R. Kosugi, T. Umeda, and Y. Sakuma, “Fixed nitrogen atoms in the SiO2 /SiC interface region and their direct relationship to interface trap density,” Appl. Phys. Lett., vol. 99, no. 18, p. 182 111, Oct. 2011. [13] Y. Xu, “Chemical state of N at SiO2 /SiC interface.” unpublished. [14] S. Dhar, A. C. Ahyi, J. R. Williams, A. K. Agarwal, and S. Ryu, “Temperature dependence of inversion layer carrier concentration and Hall mobility in 4H-SiC MOSFETs,” Mater. Sci. Forum, vol. 717-720, pp. 713–716, 2012. [15] G. L. Harris, Properties of SiC. Stevenage, U.K.: INSPEC, Inst. Elect. Eng., 1995, ser. EMIS Data reviews, no. 13. [16] C. Tin, S. Mendis, K. Chew, I. Atabaev, T. Saliev, B. Atabaev, V. Adedeji, Rusli, and E. Bakhranov, “Oxide film assisted dopant diffusion in silicon carbide,” Thin Solid Films, vol. 518, no. 24, pp. e118–e120, Oct. 2010. [17] S. Harada, S. Suzuki, J. Senzaki, R. Kosugi, K. Adachi, K. Fukuda, and K. Arai, “High channel mobility in normally-off 4H-SiC buried channel MOSFETs,” IEEE Electron Device Lett., vol. 22, no. 6, pp. 272–275, Jun. 2001. [18] F. Moscatelli, A. Poggi, S. Solmi, and R. Nipoti, “Nitrogen implantation to improve electron channel mobility in 4H-SiC MOSFET,” IEEE Trans. Electron Devices, vol. 55, no. 4, pp. 961–967, Apr. 2008. [19] S. Dhar, S.-H. Ryu, and A. K. Agarwal, “A study on pre-oxidation nitrogen implantation for the improvement of channel mobility in 4H-SiC MOSFETs,” IEEE Trans. Electron Devices, vol. 57, no. 6, pp. 1195–1200, Jun. 2010. [20] Y. Negoro, N. Miyamoto, T. Kimoto, and H. Matsunami, “Remarkable lattice recovery and low sheet resistance of phosphorus implanted 4H-SiC (1120),” Appl. Phys. Lett., vol. 80, no. 2, pp. 240–242, Jan. 2002. [21] M. A. Capano, R. Santhakumar, R. Venugopal, M. R. Melloch, and J. A. Cooper, “Phosphorus implantation into 4H-silicon carbide,” J. Electron. Mater., vol. 29, no. 2, pp. 210–214, 2000.
